Two-Step Alignment of Mixed Reality Devices to Existing Building Data
Abstract
The alignment of multi-sensory data captured with different sensors remains a challenge to this day. This framework is aimed at estimating the pose of the session origin by comparing both 3D and 2D captured data. All the poses are weighted using a number of matching characteristics and the best pose is determined.
Workflow
Bibtex
@Article{rs14112680,
AUTHOR = {Vermandere, Jelle and Bassier, Maarten and Vergauwen, Maarten},
TITLE = {Two-Step Alignment of Mixed Reality Devices to Existing Building Data},
JOURNAL = {Remote Sensing},
VOLUME = {14},
YEAR = {2022},
NUMBER = {11},
ARTICLE-NUMBER = {2680},
URL = {https://www.mdpi.com/2072-4292/14/11/2680},
ISSN = {2072-4292},
ABSTRACT = {With the emergence of XR technologies, the demand for new time- and cost-saving applications in the AEC industry based on these new technologies is rapidly increasing. Their real-time feedback and digital interaction in the field makes these systems very well suited for construction site monitoring, maintenance, project planning, and so on. However, the continuously changing environments of construction sites and facilities requires extraordinary robust and dynamic data acquisition technologies to capture and update the built environment. New XR devices already have the hardware to accomplish these tasks, but the framework to document and geolocate multi-temporal mappings of a changing environment is still very much the subject of ongoing research. The goal of this research is, therefore, to study whether Lidar and photogrammetric technologies can be adapted to process XR sensory data and align multiple time series in the same coordinate system. Given the sometimes drastic changes on sites, we do not only use the sensory data but also any preexisting remote sensing data and as-is or as-designed BIM to aid the registration. In this work, we specifically study the low-resolution geometry and image matching of the Hololens 2 during consecutive stages of a construction. During the experiments, multiple time series of constructions are captured and registered. The experiments show that XR-captured data can be reliably registered to preexisting datasets with an accuracy that matches or exceeds the resolution of the sensory data. These results indicate that this method is an excellent way to align generic XR devices to a wide variety of existing reference data.},
DOI = {10.3390/rs14112680}
}